Variable importance

disease status overall variable importance
cases overall variable importance
disease status partial dependency of lag vars by disease
cases partial dependency of variables of interest by disease

Validation

disease status direction change confusion matrix by select diseases
Disease Baseline Accuracy REPEL (Overall/New Outbreaks)
peste des petits ruminants 92% 97% / 51%
foot and mouth disease 78% 94% / 50%
african horse sickness 98% 99% / 45%
newcastle disease 83% 91% / 44%
vesicular stomatitis 90% 98% / 36%
african swine fever 93% 97% / 33%
ovine bluetongue disease 79% 90% / 26%
ovine pox disease 93% 97% / 23%
lumpy skin disease 95% 96% / 20%
highly pathogenic avian influenza 94% 93% / 19%
pleuropneumonia 96% 98% / 18%
swine vesicular disease 100% 100% / 0%
rift valley fever 97% 98% / 0%
classical swine fever 94% 96% / 0%
disease status direction change confusion matrix by taxa
Taxa Baseline Accuracy REPEL (Overall/New Outbreaks)
cervidae 73% 96% / 72%
cats 74% 96% / 71%
buffaloes 77% 96% / 63%
camelidae 76% 95% / 55%
dogs 79% 94% / 55%
sheep/goats 86% 96% / 53%
swine 87% 96% / 47%
equidae 91% 97% / 45%
cattle 86% 95% / 40%
hares/rabbits 85% 96% / 38%
birds 85% 95% / 30%
disease status direction change confusion matrix by continent
Continent Baseline Accuracy REPEL (Overall/New Outbreaks)
Americas 82% 96% / 53%
Africa 84% 95% / 49%
Asia 85% 96% / 46%
Oceania 93% 99% / 45%
Europe 87% 95% / 41%

cases model stats

## # A tibble: 6 x 4
##   model    .metric .estimator  .estimate
##   <chr>    <chr>   <chr>           <dbl>
## 1 baseline rmse    standard   129581.   
## 2 xgboost  rmse    standard   235764.   
## 3 baseline rsq     standard        0.875
## 4 xgboost  rsq     standard        0.156
## 5 baseline mae     standard     1458.   
## 6 xgboost  mae     standard     2398.
cases residuals
cases residuals by disease
cases residuals by taxa
cases residuals by continent